I have started the module, but didn't tested the sound yet. Just want to check the heat, the DC points and get the capacitors formed, and I want to say that at 100mA through the output MOSFETS (NOS Hitachi) the heat dissipation is quite high. I am working to solder bigger TO-220 heatsinks, otherwise the life of these precious devices will be really shorted.
A simple calculus will give you approximately Pdis=22V x 0,1A = 2,2W continuous dissipation. It may not sound too much, but you need big heatsinks as they will work in a closed case. If I look at your case, I can see it is not even vented.
Please have in mind, also, that Blowtorch was a closed case, but the Output devices are mounted to that case, and that act also as a heatsink.
Considering this, I would be really skeptical about removing so much heat from TO-92 devices. Of course, you may want to try, but I am just thinking of so much work you want to put into designing the PCB, just to land into a very probable bad solution... Just my opinion, of course.
EDIT: I checked the datasheets attached, and the Pmax for TO-92 devices is 1W (@25 Celsius with infinite heatsink). So the only possible solution would be to get them in TO-220 (15W) and mount them on the case
A simple calculus will give you approximately Pdis=22V x 0,1A = 2,2W continuous dissipation. It may not sound too much, but you need big heatsinks as they will work in a closed case. If I look at your case, I can see it is not even vented.
Please have in mind, also, that Blowtorch was a closed case, but the Output devices are mounted to that case, and that act also as a heatsink.
Considering this, I would be really skeptical about removing so much heat from TO-92 devices. Of course, you may want to try, but I am just thinking of so much work you want to put into designing the PCB, just to land into a very probable bad solution... Just my opinion, of course.
EDIT: I checked the datasheets attached, and the Pmax for TO-92 devices is 1W (@25 Celsius with infinite heatsink). So the only possible solution would be to get them in TO-220 (15W) and mount them on the case
Last edited:
Hi metallicus69
Thank you for support. Of course, you have a right. I did not reflect too deeply about the output. Now I see it is obvious to connect +OUT with GND. Thanks, my mistake.
What type of Hitachi transistors you used? Pair of 2SK216/2SJ79?
2,2W is rely dangerous dissipation for TO-92. Reading your I have more doubts concern to my idea to use TO-92 transistors. There is a risk - as you wrote - I put much work in the pcb design and will cause "fire" of the small transistors.
I have ordered 5 pairs of Hitachi 2SK216/2SJ79 two weeks ago. Maybe if they will arrive and will be genuine Hitachi (not fakes), I would use them instead TO-92.
You wrote: "If I look at your case, I can see it is not even vented". Where did you see my case?
I have an experience with such type of case as you wrote: transistors are fitted directly to the alu case. I have a Densen BEAT B-100 amplifier. It has whole case made of 3-5mm alu plates and transistors are fitted to the top plate. It is very efficient heat-sink. The case are never hot, even after few hours of loudly playing.
In anyway, based on your suggestions I probably serious think about TO-220.
Thank you for support. Of course, you have a right. I did not reflect too deeply about the output. Now I see it is obvious to connect +OUT with GND. Thanks, my mistake.
What type of Hitachi transistors you used? Pair of 2SK216/2SJ79?
2,2W is rely dangerous dissipation for TO-92. Reading your I have more doubts concern to my idea to use TO-92 transistors. There is a risk - as you wrote - I put much work in the pcb design and will cause "fire" of the small transistors.
I have ordered 5 pairs of Hitachi 2SK216/2SJ79 two weeks ago. Maybe if they will arrive and will be genuine Hitachi (not fakes), I would use them instead TO-92.
You wrote: "If I look at your case, I can see it is not even vented". Where did you see my case?
I have an experience with such type of case as you wrote: transistors are fitted directly to the alu case. I have a Densen BEAT B-100 amplifier. It has whole case made of 3-5mm alu plates and transistors are fitted to the top plate. It is very efficient heat-sink. The case are never hot, even after few hours of loudly playing.
In anyway, based on your suggestions I probably serious think about TO-220.
Hi flg
Thank you for other important approach of my idea. You have a right that such hot TO-92 would cause higher level of thermal noise. I see my idea has more defects than advantages. But what type of TO-220 transistors would be better use (if my ordered Hitachi 2SK216/2SJ79 will be fakes)? Any IRF510/9510 or 610/9610?
Thank you for other important approach of my idea. You have a right that such hot TO-92 would cause higher level of thermal noise. I see my idea has more defects than advantages. But what type of TO-220 transistors would be better use (if my ordered Hitachi 2SK216/2SJ79 will be fakes)? Any IRF510/9510 or 610/9610?
Hi JKoch,
I just managed to change the heatsinks and everything is fine now. I used the biggest heatsinks that could fit on the PCB.
It's my bad, I saw your posted pictures and I thought they are from your case (an intended clone of the blowtorch front panel to be used here, I thought). If you don't have another low-level preamplifier to be used in the same case, you can safely use a vented case. If it's not vented, better design your PCB with transistors mounted on the case.
I purchased some sets of pulled 2SK214/2SJ77, matched quads, from a safe source (unfortunately no longer available) and they should be identical to yours (just k216/j79 can stand bigger voltage). If I would not have these, I would probably use some IRF510/Irf9520 (or 610/9620). These have been reported as successfully used by user pinnocchio. (BIG THANKS to him, as he reported his bad experience with Hitachi, I learned that the legs are really fragile, so I cut their D and S legs short and in order to fit them in the existing PCB I used leftover terminals from the DALE resistors).
Beware that Hitachi and IRFs have different pin out (the first indication that they are not fakes), that is why you need to do the above twist. the first thing to check when you receive the Hitachi is to be sure that they have the S(source) connected to the TO-220 case . IRF's have the D(drain) internally connected to the case TAB (in the middle). Get the datasheets from the net, they are useful.
Cheers!
I just managed to change the heatsinks and everything is fine now. I used the biggest heatsinks that could fit on the PCB.
It's my bad, I saw your posted pictures and I thought they are from your case (an intended clone of the blowtorch front panel to be used here, I thought). If you don't have another low-level preamplifier to be used in the same case, you can safely use a vented case. If it's not vented, better design your PCB with transistors mounted on the case.
I purchased some sets of pulled 2SK214/2SJ77, matched quads, from a safe source (unfortunately no longer available) and they should be identical to yours (just k216/j79 can stand bigger voltage). If I would not have these, I would probably use some IRF510/Irf9520 (or 610/9620). These have been reported as successfully used by user pinnocchio. (BIG THANKS to him, as he reported his bad experience with Hitachi, I learned that the legs are really fragile, so I cut their D and S legs short and in order to fit them in the existing PCB I used leftover terminals from the DALE resistors).
Beware that Hitachi and IRFs have different pin out (the first indication that they are not fakes), that is why you need to do the above twist. the first thing to check when you receive the Hitachi is to be sure that they have the S(source) connected to the TO-220 case . IRF's have the D(drain) internally connected to the case TAB (in the middle). Get the datasheets from the net, they are useful.
Cheers!
Hi John,
Thanks for your suggestion. Unfortunately I don't have the Supertex parts, so I have to use only the Hitachi laterals.
From Kubota power supplies, I have now clean dual 30V. On the input I have J109BL/K389BL with Idss matched on all 4 channels at 10mA.
Actual DC conditions look fine (3mA through each input JFET, 100mA through each Output device) however there is still room for increase..
Can you please suggest what should I aim for in these static points?
Thank you in advance!
Thanks for your suggestion. Unfortunately I don't have the Supertex parts, so I have to use only the Hitachi laterals.
From Kubota power supplies, I have now clean dual 30V. On the input I have J109BL/K389BL with Idss matched on all 4 channels at 10mA.
Actual DC conditions look fine (3mA through each input JFET, 100mA through each Output device) however there is still room for increase..
Can you please suggest what should I aim for in these static points?
Thank you in advance!
According to this report I know I am in the ballpark at 100mA per MOSFET. If I count 40-45mA taken by Servos and input stage I can safely assume it's OK to run the preamp MOSFETS between 160/2=80mA and 240/2=120mA. I can easily measure the voltage across the series 25R Caddocks sharing the same heatsink with it's own transistor, and I have almost identical voltages (2,512V to 2,515V) and this cannot be bad at all.
The voltages across the 1k5 resistors that are the input stage loading got also to more equal values after increasing the outputs current to 100mA, and I measure between 4,317V and 4,325V across all 8 1k5 resistors on both channels.
Offset is less than 0,6mV both between OUT+ and OUT- and between each output and ground.
I have never meet so much measurement symmetry (not sure if the term is correct) in any of my previous builds (except for Paradise Phono, where I was really obsessed with matching and thermally coupling of most the transistors, having in mind MC low signals), so probably I cannot get any better with the transistors I have for this build.
Soon the moment will come to try the AC performance, and the long waited sound check. That will be only after I am sure nothing smokes and I can safely leave it on for months.
Cheers!
I have just received Hitachi 2SK216/2SJ79. I checked them ad hoc by ohmmeter only and compared them with IR IRFZ24/IRF9Z34. IRFZ24 has 2,5Mohm between D(2) and S(3) and 2SK216 has 2,0Mohm but between D(3) and S(2). Similarly IRF9Z34 has 2,75Mohm between S(3) and D(2) and 2SJ79 has 2,7Mohm but between S(2) and D(3). It seems Hitachi have changed pins to IRF according to datasheet. What other simply test I could make to confirm that the Hitachi 2SK216/2SJ79 are not fake? They have different colour of cases: N type are black and P type are green. They have printed signs as follow (is too late to make photo):
K216
7 G5
"H" (Hitachi logo)
J79
9 H5
"H" (Hitachi logo)
Someone has any advice how make simply test, please?
K216
7 G5
"H" (Hitachi logo)
J79
9 H5
"H" (Hitachi logo)
Someone has any advice how make simply test, please?
While I built the Paradise phono, I bought an Atlas DCA 75 pro, that can quickly identify the type and terminals of almost any active semiconductor device: Atlas DCA75 Pro
This handy tool can even trace the curves of the active device if using the USB connection.
It is very handy to have one and not very expensive for what it can make (you can find it also at Reichelt, ELV or on eBay). There are also cheaper alternatives (like Atlas DCA55) that can be used as well for quickly identifying the components types and terminals, and I think it's mandatory to have one in the drawer
In case you find it unnecessary, you have to build a simple circuit to test the mosfets. The normal multimeter is not enough for this task, as far as I am concerned
This handy tool can even trace the curves of the active device if using the USB connection.
It is very handy to have one and not very expensive for what it can make (you can find it also at Reichelt, ELV or on eBay). There are also cheaper alternatives (like Atlas DCA55) that can be used as well for quickly identifying the components types and terminals, and I think it's mandatory to have one in the drawer
In case you find it unnecessary, you have to build a simple circuit to test the mosfets. The normal multimeter is not enough for this task, as far as I am concerned
I have just finished the Sigma22 psu. Previously I planned to use the 100uF/50V Black Gate STD capacitors but I cannot find them in reasonable price so finally I use Rubycon 470uF/50V ZL type. They have very low ESR too. Here are schematics and pcb design. The pcb will be Au plated with double ground plane but without any solder-mask. On Monday I will send the project to manufacturer to estimate price for it.
Does this confirm that ALL lateral mosFETs have the middle pin as Source Pin?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.